I pay for ChatGPT. I manually select GPT-5.6 SOL (flagship) but it keeps silently routing me to GPT-5.5 mini. Why am I being downgraded? New chats & cache clears don't fix it. @OpenAI@thsottiaux what's going on? Anyone else? How'd you solve it? #ChatGPT#GPT5
@thsottiaux Actually, what really ruins the user experience is that the 5.6sol model degrades to the level of 5.5 mini in chat mode. Lots of users have run into this issue, and it’s seriously messing with our experience. We hope you can fix this as soon as possible.
@phiandersson@OpenAI Congrats on the new role, Philip! Sorry to hijack the thread. Several Plus users have a reproducible routing issue: gpt-5-6-thinking requests resolve server-side to gpt-5-5-mini. Support hasn’t helped (Case 13585326). Could you point us to the right ChatGPT engineer?
Anyone else seeing this with ChatGPT Chat mode the past few days?👀
When I select High (or any higher reasoning effort), it silently routes to GPT-5.5-mini.
But if I stay on Instant, it correctly runs as GPT-5.6 Sol.
Work mode is completely fine and stays on real 5.6 Sol.
The issue only shows up in Chat’s higher reasoning modes.✨
Is this happening to anyone else? Especially on mobile app?
Would love to know if it’s widespread or just some accounts.
#ChatGPT
@jamesfzhang@din0s_@jamesfzhang Could this be related? Plus requests gpt-5-6-thinking, but telemetry resolves gpt-5-5-mini with no limit notice. It affects multiple browsers and fresh accounts; support hasn’t helped. Could you verify or route Case 13585326 to the model-routing team?
@michpokrass@igormaka@simonw@michpokrass Hi Michelle — I’m seeing a server-side routing issue: Plus requests gpt-5-6-thinking, but telemetry resolves gpt-5-5-mini, with no limit notice. Support hasn’t been able to help. Could you please forward Case 13585326 to the model-routing team? @ChatGPTapp
@babayagatwt You gotta be ragebaiting, ive seen at least 500 different people complaining about openai support ignoring them and that so many people are getting routed to GPT 5.5-mini for absolutely no reason, but yeah sure really great support
@thsottiaux Have you also investigated the problem beeing routed from 5.6 Sol high and extra high down to 5.5 mini? I just noticed changes in Sol's response and checked it out and saw the model that was solving was 5.5 thinking mini.
Sol 5.6 has been "nerfed" to the point that I clearly prefer Qwen 3.8 27B over it. OpenAI has made it genuinely unusable for anything serious.
Is anyone having a similar experience?
@thsottiaux Tibo please fix that people are being routed to gpt 5.5 mini we are using gpt 5.6 sol but no matter what we pick we get 5.5 mini at least give us a reason why
@thsottiaux bro please fix it i keep being routed to gpt 5.5 mini even with using gpt 5.6 sol its so unusable and the discord is saying it too im not the only one i cant survive the weekend without big chatgpt